Transparency and controllability in user interfaces that
adapt during run-time
Matthias Peissner
Fraunhofer IAO
Nobelstr. 12
70569 Stuttgart, Germany
+49 711 970 2311
[email protected]
ABSTRACT
This position paper presents research issues and solutions
from the current European research project MyUI 1. MyUI
provides a framework and a technical infrastructure for
adaptive user interfaces to increase the accessibility of ICT
products. After an introduction of the MyUI approach
which relies on the run-time composition of design patterns,
we highlight our contribution to increasing the transparency
and controllability of self-adaptive user interfaces. Two
dedicated categories of MyUI design patterns have been
developed: (1) Adaptation Rendering Patterns describe
solutions for transparent transitions by animations; (2)
Adaptation Dialogue Patterns cover dialogue strategies to
put the user into control of the system-initiated adaptations.
A current user study evaluates the effectiveness of the
proposed patterns. The results will be available before the
workshop.
Keywords
Adaptive user interfaces, design patterns, transparency,
controllability, accessibility
INTRODUCTION
Self-learning and self-adaptive user interfaces offer great
opportunities for accessibility and improved usability for a
broad range of diverse users. However, they also pose
significant challenges in terms of usability, trustworthiness
and acceptability. Consistency is a main quality of usable
and reliable user interfaces. System-initiated adaptations
will lead to changes in the appearance of the user interface.
In many cases, adaptations will occur in situations of
interaction problems. Therefore, it is important that changes
triggered by adaptation processes do not lead to further
disorientation but support the user in recovering from an
experienced interaction problem. A second important topic
for user interface adaptations is their transparency. The
users must be able to recognize changes in the user
interface and they should also be able to understand why
the change has been made. Finally, intelligent and
automatic adaptations must not lead to situations where the
end user feels like losing control over the user interface.
1
This work was funded by the European Union MyUI
project under grant FP7-ICT-248606
Thomas Sellner
Fraunhofer IAO
Nobelstr. 12
70569 Stuttgart, Germany
+49 711 970 2347
[email protected]
MyUI recognizes that the design of the adaptation process
is a major issue for the usability and acceptability of
adaptive user interfaces. MyUI addresses this important
field with dedicated Adaptation Patterns. They inform the
users about intended or ongoing adaptations, and provide
the users with effective control mechanisms. This paper is
structured as follows: the next section provides a short
summary of some related work. Then, the overall MyUI
system is introduced. In the main part, we describe the
MyUI adaptation patterns which aim at increasing the
transparency and controllability. Finally, a user study for
evaluating the proposed solutions is outlined and open
research issues are summarized.
RELATED WORK
For the personalization of user interfaces, mainly two
approaches exist. Adaptable user interfaces provide the user
with mechanisms to actively customize the interface,
whereas adaptive user interfaces initiate and perform
dynamic adjustments autonomously (cf. [4], [9]).
Both approaches have their pros and cons. The most
important advantage of adaptable systems is that the users
are in total control of the individual appearance of their user
interface. This supports the understandability and
traceability of modifications from the user’s perspective (cf.
[7]). However, this advantage is at the same the main a
shortcoming of adaptable user interfaces – especially when
personalization aims at increasing the accessibility. Users
with disabilities and lower levels of ICT literacy would
benefit most from personalized user interfaces. They often
have severe problems with standard configurations.
Customization dialogues, however, are a significant barrier
- even for abled users [10].
On the other hand, adaptive user interfaces are often
blamed for confusing the users with unpredictable
adaptation behaviors which cannot be understood nor
controlled by the users [14]. Mechanisms to overcome the
lacks of transparency and controllability in adaptive user
interfaces have received relatively little attention in the
literature. In a recent paper, Dessart et al. describe a first
attempt to improving the transparency of run-time
adaptations by animated transitions. They propose a
taxonomy of adaptation categories and suitable transitions
[3]. For the problem of controllability, not much specific
guidance is available. Findlater and McGrenere provide an
overview of some empirical studies which compare
adaptable and adaptive user interfaces. On the basis of their
own study on personalized menus they conclude that a
mixed-initiative design which combines adaptable and
adaptive mechanisms in one system will be the best to
satisfy diverse users [6]. MICA is a good example for a
modern mixed-initiative system which recommends
customizations but leaves the decision to the user [1].
However, adaptive user interfaces that aim at accessibility
improvements might need approaches where the system
takes a more proactive role.
1. User Interface Parameterization: In a first step,
general user interface settings are defined to cover
individual user needs and current context conditions. This
first step can be regarded as “translating” information
about the user and context (from the user profile) and the
used device (from the device profile) to user interface
settings and storing them in the user interface profile.
OVERVIEW OF THE MYUI SYSTEM
3. User Interface Generation and Adaptation: The
selected user interface components are rendered to an
individual user interface. Decisions about necessary
adaptations to the currently displayed user interface are
made and executed.
MyUI provides individualized user interfaces which are
accessible to a broad range of users. The conceptual
framework of MyUI contributes to the further development
in the field of adaptive user interfaces by focusing on the
following aspects which are often not sufficiently covered
by existing solutions (e.g. [2], [8], [13]):

Generic framework for manifold adaptations: The
MyUI technical framework allows for adaptations to
diverse user needs, different devices2 and changing
context conditions. Therefore, MyUI user interfaces adapt
their presentation formats and modalities, the interaction
mechanisms and the navigation paths.

Modular, extensible and open: A modular approach is
taken to manage the huge amount of possible user
interface solutions. For practical reasons, the extensibility
of the modular approach is regarded important to support
a quick start with a manageable subset of design solutions
and later extensions. Modularity and extensibility are
achieved by a design patterns approach to adaptive user
interfaces. The MyUI design patterns repository is
publically available [11]. Thus, the underlying adaptation
rules can be reviewed, refined and extended by other
experts in the field.

Self-learning and adapting during use: Not only
before but also during the interaction, the MyUI system is
collecting information about the user and updates the user
profile accordingly. In order to cover dynamic user profile
changes, the MyUI framework supports run-time
rendering and run-time adaptations of the user interface.

Transparent and controllable: To assure high levels of
usability and user acceptance, the MyUI approach strives
to provide mechanisms which help the users to recognize
and understand user interface adaptations [12].
Three-stage process of user interface generation and
adaptation
MyUI adaptive user interfaces are generated and adapted in
a three-stage process (see Figure 1):
2
The current MyUI application prototypes run on a webbased iTV platform. The MyUI technical framework,
however, is generic and not restricted to iTV.
2. User Interface Preparation: On the basis of the
current interaction situation and user interface profile,
suitable display and control elements are selected for an
optimal support of the individual user in the current state
of the application.
In a permanent process, relevant events from the interaction
are fed back to the MyUI User and Context Management
Infrastructure to keep the user profile up-to-date. Profile
updates trigger the three-stage adaptation process again.
Figure 1 MyUI
adaptation
user
interface
generation
and
MyUI design patterns
User interface generation and adaptation is done by the
composition of multimodal user interface design patterns.
Individual accessibility is achieved by combining design
patterns that suit for a certain user and context. Adapting
the user interface means switching from one design pattern
of a cluster to another pattern of the same cluster which is
hypothesized to fit best to the individual user’s needs.
The MyUI design patterns repository includes different
categories of patterns. Each pattern type fulfills distinct
functions in the MyUI adaptation framework (see Figure 1):
Device-specific patterns create a device profile on the basis
of primitive device features as provided by the used I/O
devices. Changes in the device setup will lead immediately
to updating the device profile.
Individualization patterns define global user interface
settings to fit individual user needs and specific context
conditions, e.g. big font size for users with low vision or
iconographic presentation for users with problems in
language reception. As the available knowledge about the
user and context can change during the interaction,
individualization patterns play an important role for user
interface adaptations during use.
These first two pattern types work on a general level of the
user interface. Together, they are responsible for setting and
adjusting global variables in the user interface profile. An
additional input to the user interface profile comes from the
customization profile which allows for corporate-, projector brand-specific customizations.
Interaction patterns provide suitable user interface
components for a current interaction situation as specified
in the abstract application interaction model (AAIM), e.g. a
list element for an interaction situation in which a user can
select from a set of options. For each interaction situation, a
bundle of different interaction design patterns exists. They
differ in appearance or input mechanisms to support
different user needs. The selection of the best suitable
interaction pattern from a current bundle is done on the
basis of specific variables of the user interface profile.
User interface elements are the building blocks for the
interaction patterns. While interaction patterns can be
regarded as components to support a given interaction
situation, the user interface elements provide generic
primitives required to compose the interaction patterns.
Adaptation patterns cover the dynamics of the adaptation
processes, i.e. they define the mechanisms of switching
from one instance of a user interface to another. Adaptation
patterns are described in more detail in the next section.
between the user interface before and after the adaptation.
Animated transitions shall draw the end user’s attention to
the screen areas where adaptations occur and shall help
them to understand that and how the new user interface is a
modification of the former user interface. When, for
example, an increased font size results in hiding menu
options which were directly available before the adaptation,
an animation can communicate to the user that the hidden
options are now available via the »more« button.
Adaptation dialogue patterns
Adaptation dialogue patterns specify the dialogue between
the user interface and the MyUI user in the course of an
adaptation. This dialogue typically includes a notification
and interaction options for the user to influence the
system’s adaptation behavior. In MyUI, two types of
adaptation dialogue patterns are distinguished:

System-initiated adaptation dialogue patterns are
triggered by the system.

User-initiated adaptation dialogue patterns describe
customization dialogues in which the user modifies the
user profile or the user interface profile.
The latter make MyUI a mixed-initiative system with
adaptive and adaptable components. More interesting,
however, are system-initiated adaptation dialogue patterns.
They aim at increasing the usability and acceptability by
making system-initiated adaptations more transparent and
controllable. The MyUI patterns repository includes the
following three system-initiated adaptation dialogue
patterns:

MYUI ADAPTATION PATTERNS
Adaptation patterns are a major concept in the MyUI
adaptation process. Their main purpose is to ensure high
levels of transparency and controllability for the user in
situations of adaptations during use.
Explicit Confirmation before Adaptation: Before
performing the adaptation the system requests the user to
explicitly accept or reject the adaptation. The user’s
decision is supported by providing a preview of the
adaptation effect (Figure 2).
User interface adaptations are executed by the MyUI
Adaptation Engine. The adaptation engine recognizes
mismatches of the user interface components currently
displayed at the user interface and the components currently
selected by the user interface preparation process. A
mismatch triggers a run-time adaptation. Adaptations are
performed by two types of MyUI adaptation patterns:


Adaptation rendering patterns specify the graphical
rendering process to smoothly but obviously switch from
one user interface instance to another, e.g. animated
transitions to grow small fonts to bigger fonts
Adaptation dialogue patterns specify the interaction
dialogue which takes place around the actual adaptation
to make sure that the user is aware of the adaptation and
can control the system’s adaptation behavior.
Adaptation rendering patterns
MyUI adaptation rendering patterns make extensive use of
animations. They support orientation by creating continuity
Figure 2 »Explicit Confirmation before Adaptation«:
Dialogue box with preview (UI concept sketch)

Explicit Confirmation after Adaptation: The adaptation
is triggered and performed automatically. When the
adaptation has been finished, a dialogue box asks the user
if the changes shall be kept or undone.

Automatic Adaptation with Implicit Confirmation: The
adaptation is triggered and performed automatically.
While rendering the adaptation, the system provides an
icon-based notification about the on-going adaptation in a
dedicated adaptation area on the screen. When the
adaptation is finished the end user can undo the
adaptation via a button in the adaptation area.
User study on adaptation dialogue patterns
We assume that the effectiveness and acceptability of the
three adaptation dialogue patterns depend on a subjective
cost-benefit judgment of the specific adaptation.
Predominant costs might be associated with a higher wish
for user control and therefore, a preference for explicit
confirmations. However, in situations where the benefits
outweigh the costs, user might prefer automatic adaptations
because of their higher comfort of use. Findlater and Gajos
also point to costs of adaptations, e.g. caused by incorrect
adaptations or decreased aesthetics of the adapted UI [5]. In
our case, adaptations can cause further costs, for example,
when the adapted user interface requires additional
interaction steps, e.g. increasing the font size requires extra
scrolling. Benefits occur when an adaptation can actually
increase the accessibility by eliminating a barrier of use.
Besides these situational factors, also personal traits might
influence the subjective cost-benefit ratio and the
preference of one of the adaptation dialogue patterns. Users
with lower levels of ICT literacy or a higher need for
security might tend to over-estimate the costs and therefore,
prefer explicit confirmations.
We are currently preparing a user study addressing the
following questions:



Which adaptation dialogue patterns are most effective
in terms of transparency, controllability and acceptance?
Which pattern is preferred by the users?
Do these measures differ in different cost-benefit
conditions and for different users?
The results will be available by end of March 2012. They
shall help us to design the mechanism for selecting the most
appropriate adaptation pattern for different situations and
user profiles.
REFERENCES
1. Bunt, A., Conati, C., & McGrenere, J. (2009). A MixedInitiative Approach to Interface Personalization. AI
Magazine 30(4).
2. Coelho, J., Duarte, C., Biswas, P. & Langdon, P.
(2011). Developing accessible TV applications. In The
proceedings of the 13th international ACM
SIGACCESS
conference
on
Computers
and
accessibility (ASSETS '11). ACM, New York, 131-138.
3. Dessart, C.-E., Motti, V. G. & Vanderdonckt, J. (2011).
Showing user interface adaptivity by animated
transitions. In Proceedings of the 3rd ACM SIGCHI
symposium on Engineering interactive computing
systems (EICS '11). ACM, New York, 95-104.
4. Dieterich, H., Malinowski, U., Kühne, T. & SchneiderHufschmidt, M. (1993). State of the Art in Adaptive
User Interfaces. In: M. Schneider-Hufschmidt, T. Kühne
& U. Malinowski (Eds.): Adaptive User Interfaces:
Principles and practice. Amsterdam: North-Holland, 13–
48.
5. Findlater, L. & Gajos, K.Z. (2009). Design Space and
Evaluation Challenges of Adaptive Graphical User
Interfaces. AI Magazine 30(4).
6. Findlater, L. & McGrenere, J. (2004) A comparison of
static, adaptive, and adaptable menus. In Proceedings
SIGCHI Conference on Human Factors in Computing
Systems (CHI’04), New York: ACM, 89-96.
7. Fischer, G. (1993). Shared knowledge in Cooperative
Problem-Solving Systems – Integrating Adaptive and
Adaptable Components. In: M. Schneider-Hufschmidt,
T. Kühne & U. Malinowski (Eds.): Adaptive User
Interfaces: Principles and practice. Amsterdam: NorthHolland, 49–68.
8. Gajos, K. Z., Weld, D. S., & Wobbrock, J. O. (2010).
Automatically generating personalized user interfaces
with Supple. Artificial Intelligence 174, 12-13. 910–
950.
9. Kobsa, A., Koenemann, J., & Pohl, W. (2001).
Personalised hypermedia presentation techniques for
improving online customer relationships. The
Knowledge Engineering Review, Vol. 16:2, 111–155.
10. Mackay, W. E. (1991). Triggers and barriers to
customizing software, in: S. P. Robertson, G. M. Olson
& J. S. Olson (Eds.). Proceedings of the SIGCHI
Conference on Human Factors in Computing Systems
(CHI ’91), New York: ACM. S. 153–160.
11. MyUI Design Patterns Repository.
http://myuipatterns.clevercherry.com
Available
at
12. Peissner, M., Schuller, A. & Spath, D. (2011). A design
patterns approach to adaptive user interfaces for users
with special needs. In: J.A. Jacko (Ed.): HumanComputer Interaction, Part I, HCII 2011, LNCS 6761,
Berlin: Springer-Verlag, pp. 268–277.
13. Stephanidis, C., Paramythis, A., Sfyrakis, M., Stergiou,
A., Maou, N., Leventis, A., Paparoulis, G. &
Karagiannidis, C. (1998). Adaptable and adaptive user
interfaces for disabled users in the AVANTI project, in:
IS & N 98 Proceedings (1998), Springer-Verlag, 153166.
14. Weld, D., Anderson, C., Domingos, P., Etzioni, O., Lau,
T., Gajos, K. & Wolfman, S. (2003). Automatically
personalizing user interfaces. In Proceedings of the 18th
international joint conference on Artificial intelligence
(IJCAI'03). Morgan Kaufmann Publishers Inc., San
Francisco, CA, USA, S. 1613-1619.
The columns on the last page should be of equal length.